Time Series & Forecasting, Winter, 2018
The objective of the course is to learn new tools and models for time series data while paying particular attention to forecasting. We will start with different types of time series we may encounter in practice. Then we will consider the simple smoothing techniques which help us to decompose a time series into additive or multiplicative components. Next we discuss the general ideas of forecasting, types of forecasts and popular performance measures. Building upon this ideas we will move towards simple forecasting methods based on exponential smoothing. Within the next block of lectures, we will discuss ARMA type models both from theoretical and practical perspectives. Here we will concentrate on estimation, model selection, and forecasting. Additionally, we will consider extensions to seasonal models and models with conditional heteroscedasticity. Within the next block, we will learn how to model specific time series, such as time series with many zeros, how to deal with structural changes and how to aggregate forecasts.
- Typical patterns in time series data
- Time series decomposition
- Types of forecasts
- Performance measures
- Forecast averaging
- Forecasting using regression
- ARMA modelling
- Autocorrelation function
- Theoretical time series
- Estimation and order selection
- ARCH/GARCH: models with conditional volatility
- Basic models for multivariate time series data
- Forecasting special data
- Binary data
- Time series with many zeros
- Structural breaks
- Forecast aggregation
Python or R (preferable), Theory of Probability, Statistics.
Dr. Yarema Okhrin
Affiliation: University of Augsburg, Germany
After my PhD and PostDoc in Statistics and Econometrics at the European University Viadrina, Germany I spent two years as Assitant Professor in Econometrics at the University of Bern, Switzerland. Since 2010 I have been holding the chair of Statistics at the University of Augsburg, Germany.
Fields of interests: financial econometrics, multivariate and high-dimensional statistical analysis, dependence modelling, environmental statistics.